Research On Detection Techniques Of Ultra-wide Band Radar Targets

Ultra-Wide Band (UWB) radar possesses a very broad application prospect in the fields of target identification and imaging, etc., since the radar usually has outstanding performance of range resolution. However, the Doppler frequency of UWB radar target with constant radial velocity varies with the frequencies of UWB signal, i.e. Doppler dispersion phenomenon. This phenomenon becomes more prominent than narrow band radar, and makes it difficult to estimate velocity of moving target for UWB radar by using the conventional signal processing methods suitable for narrow radar. Besides, for moving target with radial velocity, the modulation UWB signals, such as multi-carrier frequency and linear frequency modulation (LFM), etc., have range-Doppler coupling, which causes that the range profile moves and spreads seriously along range dimension. It is found that the capability of high range resolution of UWB radar is greatly restricted. This thesis focuses on target detection of range and velocity profiles for UWB radar, including1. Analysis on Doppler dispersion characteristicThe echo of UWB radar is modeled by using Lorentz transformation and wave4-vector with three-dimensional space and time from the point of electromagnetic wave progagation. Doppler dispersion effect on the performance of matched filter output is discussed for UWB radar based on this echo model. In addition, the condition suitable for matched filter is quantified, i.e. when the Doppler dispersion product BT(2v/c) is smaller than1, the echo can be considered as non-stretch and the tranditional processing of matched filter is fit for this case, and vice versa counter, where B, T, v and c are signal bandwidth, pulsewidth, relative radial velocity and light speed, respectively.2. Range profile and velocity estimation of target for UWB radar based on stepped frequency pulse trains signalStaggering of pulse repetition interval (PRI) varying with subband is implemented for Doppler dispersion caused by the stepped frequencies of stepped frequency pulse trains (SFPTs) signal to solve the performance degradation of range profile. The staggering not only completely eliminates the effect of Doppler dispersion on synthetic range profile, but also relieves the restriction of the number N of the pulse trains. Then, when target velocity is estimated via fast Fourier transform processing, a quantized error is caused. Therefore, iterative dimidate approaching method (IDAM) is proposed in order to improve the calculated velocity precision. This method can improve the accuracy of N times by log2VN iterative processing, and the number of range motion bins caused by calculated velocity error is smaller than a half of range bin. However, the pulses number of SFPTs signal is larger than conventional stepped frequency signal since each subband includes multiple pulses with the same carrier frequency, which not only causes a moving target moving along multiple range cells, but also a constant velocity is hardly maintained. So a multiple signal classification (MUSIC) method is ultilized to estimate high resolution radial velocity on the basis of SFPTs signal. Compared with IDAM, this method still has better performance of velocity estimation even if in the case of low signal-to-noise ratio (SNR). In addition, the method usually acquires less number of pulses in each subband if the requirement, i.e. the number of pulses in each subband is larger than the number of targets moving with different radial velocities, can be met. Finally, with the emerging of high-speed moving vehicles, this thesis further study resolving velocity ambiguity for target moving at high velocity based on SFPTs signal.3. Range and velocity estimation of moving target for LFM UWB radarFor a moving target with the Doppler dispersion product is smaller than1for LFM UWB radar, two-dimensional projection discrete Fourier fransform and velocity compensation (2-DPDFTVC) technique is proposed. In this technique, two-dimensional projection discrete Fourier transform (2-DPDFT), used for direction of arrival (DOA) estimation in the case of wideband signal, is referenced to obtain high resolution estimated velocity. Then, this method can further get high resolution range profile after the corresponding compensation with the calculated velocity. This thesis derives and quantifies the range of velocity according to the Doppler dispersion product being smaller than1. When the Doppler dispersion product is larger than1, state space technique is used to obtain range profile and velocity with high resolution. Then, Cramer-Rao lower bound (CRLB) of the estimated parameter is derived. The CRLB is compared with root mean square error (RMSE) obtained through Monte Carlo duplicate test in order to get the statistical performance of the parameter. State space method possesses fine estimation performance in the condition of low SNR. Besides, the difference between state space method and wavelet transform is that the former does not have admissibility restriction because the former only acquires impulse response of target in frequency domain and does not consider the special transimitted signal waveform.4. UWB radar clutter and its suppressionA method of frequency splitting and subbands synthesis, combining with facet physical model, is proposed to construct UWB radar clutter model. This model not only reflects the feature varing with UWB frequency, but also reports the effects of many detail factors, such as polarization, electromagnetic properties of ground, landforms bump and vegetation cover, etc., on the clutter model. Maximum likelihood estimator (MLE) and Kolmogorov-Smirnov test are utilized to obtain probability statistical model having fine goodness of fit with the histogram of clutter simulation data and the model’s parameters, respectively. The corresponding simulation results show that the probability statistical model of UWB clutter has the characteristics of the model varying with the parameters of radar system and clutter, large probability near0radar cross section and long tail. On the basis of the clutter model and the related simulation, this dissertation establishes an echo model mixing with the information of target and clutter for UWB radar. Two methods, called noncoherent moving target indication (NMTI) and filtering directly in Doppler-domain, are introduced to reject UWB radar clutter. Rang profiles are compared before and after NMTI and filtering directly in Doppler-domain processing to demonstrate the effect of clutter suppression, their restriction and the problems for improvement.